SCHEDULED AGENTS

Put your agents on a schedule. Wake up to finished work.

Triage issues every morning. Check dependencies every week. Run a long refactor overnight. Pick an agent, write the prompt and set a schedule. Your computer starts the work on time, even with your browser closed.

  • Once, hourly, daily, weekly or cron
  • Runs with your browser closed
  • Resume any run

What are Scheduled Agents?

Scheduled Agents is TermSquad’s scheduler for AI coding agents. It runs an agent or a Squad on your TermSquad Computer at the times you choose and keeps a history of every run, so you can review the result or resume the conversation.

Why schedule agents

The routine work.
Done before you start.

Much of the most useful agent work is routine: a morning triage, a nightly test sweep, a weekly dependency check. On an always-on computer, that work happens on time, without you remembering to start it or keeping a laptop awake.

Every morning

Start the day triaged.

Have an agent read new issues, label them and draft a plan, so you open your laptop to a clear list instead of a pile.

Daily or weekdays
Every hour

Keep an eye on the build.

Ask an agent to check CI, investigate failures and fix flaky tests while you focus on something else.

Hourly runs
Every weekend

Hand bigger jobs to a Squad.

Schedule a Squad with implementer and reviewer roles for dependency upgrades or cleanups that need a second pair of eyes.

Squads with up to 8 collaborators

Your computer, your agents. Each run uses the agents, repositories and credentials already on your TermSquad Computer, in a fresh conversation.

How Scheduled Agents works

Pick the work.
Pick the time.

Open Schedule in the app sidebar, below Files. The next five run times appear before you save, so you know exactly what will happen.

  1. 01

    Create a task.

    Open Schedule in the app and create a task: choose the workspace, folder, agent or Squad, and write the prompt.

  2. 02

    Set the schedule.

    Choose when it runs: once, hourly, daily, weekly or with a cron expression, in your timezone and with a timeout.

  3. 03

    Review every run.

    Your computer starts each run on time, in a fresh conversation. Review the result in Execution history, resume the conversation or run it again.

When an agent needs you. With YOLO mode off, a run that asks for approval stops with Needs attention and keeps the conversation, ready for you to resume.

Built for unattended work

Clear results.
Clear limits.

One scheduled run at a time on each computer, a timeout on every run and no automatic retries. A run is complete only when the agent reports back.

Scheduled Agents includes
  • Once, hourly, daily, weekly or cron schedules
  • Your timezone, with the next five run times shown
  • A single agent or a Squad with implementer and reviewer roles
  • Your choice of workspace, folder, agent, model and effort
  • A timeout from 1 minute to 24 hours
  • An optional YOLO mode per task, off by default
  • Execution history with status, duration and result
  • Resume, Run again and Run now actions

Runs that are more than ten minutes late are skipped and recorded, so downtime never floods your computer with a backlog.

Frequently asked questions

Set it once.
Know what to expect.

How scheduled runs start, what happens when an agent needs you and how to read the results.

Explore the documentation
Does my browser or laptop need to be open?

No. Scheduled runs start on your TermSquad Computer. The computer needs to be running; your browser and laptop can be closed.

Can I schedule a Squad instead of a single agent?

Yes. A task can use one agent or a Squad with implementer and reviewer roles. A scheduled Squad can run up to eight collaborators at once.

What happens if an agent asks for approval?

With YOLO mode off, the default, the run stops with the status Needs attention and keeps the conversation, so you can resume it and answer. YOLO mode turns on the agent’s own option to skip approvals for that task only. It does not change Linux permissions.

Can several scheduled tasks run at the same time?

Each computer runs one scheduled task at a time, and the next one waits its turn. Your manual agent sessions are separate and keep working. Runs are not retried automatically.

How do I know whether a run succeeded?

A run is marked complete only after the agent sends its report. Execution history shows the task, agent, workspace, start time, duration, status and result. The result is the agent’s own report, so review its work as you would any agent output.

What if my computer was off at the scheduled time?

A run that is more than ten minutes late is skipped and recorded in the history instead of starting late. That way, downtime never floods your computer with a backlog of old runs.

Work that happens on time

Put the routine on autopilot.
Review it with coffee.

Your agents. Your schedule. Your always-on cloud computer.

AI subscriptions and usage are not included.